EGU24-8516, updated on 08 Mar 2024
EGU General Assembly 2024
© Author(s) 2024. This work is distributed under
the Creative Commons Attribution 4.0 License.

Pluvial Flooding Risk Mitigation: a machine learning approach for optimal management of an urban drainage system

Sabrina Lanciotti1, Leonardo Alfonso2, Elena Ridolfi1, Fabio Russo1, and Francesco Napolitano1
Sabrina Lanciotti et al.
  • 1Sapienza, University of Rome, Dipartimento di Ingegneria Civile Edile e Ambientale (DICEA), Rome, Italy (
  • 2IHE Delft Institute for Water Education, Department of Hydroinformatics and Socio-Technical Innovation, Delft, the Netherlands

According to the Intergovernmental Panel on Climate Change (IPCC), the variability of extreme rainfall events is increasing in many locations. The continuous expansion of urban areas makes urban flooding more common, thus increasing the need for improved management of drainage systems in large cities. Urban pluvial flooding (UPF) occurs when surface runoff cannot be efficiently conveyed into the drainage system, due to intense rainfall events exceeding the capacity of stormwater drainage systems, or due to inlets' poor maintenance which are often either partially or fully blocked. Many drainage systems may not be efficient due to outdated design approaches that do not consider these aspects. Therefore, there is a need to improve the design of structures and to prioritize risk adaptation and mitigation strategies to build resilient cities against the effects of pluvial flooding. During extreme rainfall events generating pluvial flooding, discharges exceeding the sewer system capacity are diverted by sewer overflows. For this reason, the objective of this work consists of defining a methodology to determine the optimal management strategy to mitigate sewer overflows using machine learning techniques. Here we simulate pluvial flooding within a large urban area by using the freely available Storm Water Management Model EPA-SWMM, based on a detailed reproduction of the geometric characteristics of a branch of the drainage network in a large city. By simulating the different propagation effects of synthetic hyetographs in the pipelines and artefacts, the dynamic operating conditions of the actual network are performed using machine learning techniques, by applying Python to the SWMM data model (PySWMM). The project, which is ongoing, thus aims at the optimal management of the combined overflow devices of a sewer system through their real time control during a flood event to mitigate pluvial flooding risk in urban areas.

How to cite: Lanciotti, S., Alfonso, L., Ridolfi, E., Russo, F., and Napolitano, F.: Pluvial Flooding Risk Mitigation: a machine learning approach for optimal management of an urban drainage system, EGU General Assembly 2024, Vienna, Austria, 14–19 Apr 2024, EGU24-8516,, 2024.